混流式水轮机神经网络模型非线性仿真

针对水轮机非线性特性难以准确描述及混流式水轮机调节系统(FTGS)非线性仿真复杂性，利用前馈神经网络强大的非线性逼近能力，建立混流式水轮机神经网络模型(FTNNM)。描述流量和效率特性的FTNNM采用Levenberg—Marquardt算法进行离线训练，收敛速度快、模型精度高。训练后的FTNNM作为一个非线性环节和其他模块构成了在MATALAB的SIMUUNK环境下的非线性仿真系统。该系统实现了各种运行工况的非线性仿真，并能快速、准确地得到系统以及混流式水轮机内部各种参数的变化规律。为混流式水轮机调节系统(FTGS)控制策略的研究提供了一个良好的基础。

Neural Network Non-Linear Emulation Modeling for Francis Turbine

Aimed at the problem that it is difficult to describe the non linear properties of water turbines and the complicated non linear properties of the Francis turbine governing system (FTGS), a Francis turbine neural network model (FTNNM) is established by using the strong approaching ability of forward feedback neural network. In the model, the properties of discharge and efficiency properties are described in the off line train by adoption of Levenberg Marquardt algorithm, which has the effect of quick convergence and high precision. After trained, the model can be used as a non linear section to compose a non linear emulation system together with other modules under the circumstances of SIMULINK of MATALAB. This system formed can realize non linear emulation at different operation conditions, and can find the change rules of internal parameters of Francis turbine rapidly and accurately. Thus, the system can provide a good foundation for the study on the control strategies for Francis turbine governing system (FTGS).